Note that this test isn’t the same as raw converter performance, because there is the overhead of JS bindings, which is very significant. I have seen 9.1 GB/s for English UTF-8 decode in a non-browser test harness without extra copying while this benchmark shows 1.3 GB/s on the same computer for the same converter implementation.
Not started
What | Language | Encoding | MiB/s | Mi u16s / s | ns / iteration |
---|